Corrupted by algorithms?: How AI-generated and human-written advice shape (dis)honesty

Margarita Leib; Nils Köbis; Rainer  Michael Rilke; Marloes  Hagens; Bernd  Irlenbusch

doi:10.1093/ej/uead056

Corrupted by algorithms? How AI-generated and human-written advice shape (dis)honesty

Margarita Leib, Nils Köbis, Rainer Michael Rilke, Marloes Hagens, Bernd Irlenbusch^*

^*Corresponding author for this work

Social Psychology

Research output: Contribution to journal › Article › Scientific › peer-review

140 Downloads (Pure)

Abstract

Artificial Intelligence (AI) increasingly becomes an indispensable advisor. New ethical concerns arise if AI persuades people to behave dishonestly. In an experiment, we study how AI advice (generated by a Natural-Language-Processing algorithm) affects (dis)honesty, compare it to equivalent human advice, and test whether transparency about advice source matters. We find that dishonesty-promoting advice increases dishonesty, whereas honesty-promoting advice does not increase honesty. This is the case for both AI- and human advice. Algorithmic transparency, a commonly proposed policy to mitigate AI risks, does not affect behaviour. The findings mark the first steps towards managing AI advice responsibly.

Original language	English
Article number	uead056
Journal	Economic Journal
Volume	uead056
DOIs	https://doi.org/10.1093/ej/uead056
Publication status	Accepted/In press - 2023

Access to Document

10.1093/ej/uead056Licence: CC BY

SP_Leib_Corrupted by Algorithms_TEJ_2023_PreprintSubmitted manuscript, 2.8 MBLicence: CC BY

Cite this

@article{ff625847700342afbba486496f775e8e,

title = "Corrupted by algorithms?: How AI-generated and human-written advice shape (dis)honesty",

abstract = "Artificial Intelligence (AI) increasingly becomes an indispensable advisor. New ethical concerns arise if AI persuades people to behave dishonestly. In an experiment, we study how AI advice (generated by a Natural-Language-Processing algorithm) affects (dis)honesty, compare it to equivalent human advice, and test whether transparency about advice source matters. We find that dishonesty-promoting advice increases dishonesty, whereas honesty-promoting advice does not increase honesty. This is the case for both AI- and human advice. Algorithmic transparency, a commonly proposed policy to mitigate AI risks, does not affect behaviour. The findings mark the first steps towards managing AI advice responsibly.",

author = "Margarita Leib and Nils K{\"o}bis and {Michael Rilke}, Rainer and Marloes Hagens and Bernd Irlenbusch",

year = "2023",

doi = "10.1093/ej/uead056",

language = "English",

volume = "uead056",

journal = "Economic Journal",

issn = "0013-0133",

publisher = "Wiley-Blackwell",

}

TY - JOUR

T1 - Corrupted by algorithms?

T2 - How AI-generated and human-written advice shape (dis)honesty

AU - Leib, Margarita

AU - Köbis, Nils

AU - Michael Rilke, Rainer

AU - Hagens, Marloes

AU - Irlenbusch, Bernd

PY - 2023

Y1 - 2023

N2 - Artificial Intelligence (AI) increasingly becomes an indispensable advisor. New ethical concerns arise if AI persuades people to behave dishonestly. In an experiment, we study how AI advice (generated by a Natural-Language-Processing algorithm) affects (dis)honesty, compare it to equivalent human advice, and test whether transparency about advice source matters. We find that dishonesty-promoting advice increases dishonesty, whereas honesty-promoting advice does not increase honesty. This is the case for both AI- and human advice. Algorithmic transparency, a commonly proposed policy to mitigate AI risks, does not affect behaviour. The findings mark the first steps towards managing AI advice responsibly.

AB - Artificial Intelligence (AI) increasingly becomes an indispensable advisor. New ethical concerns arise if AI persuades people to behave dishonestly. In an experiment, we study how AI advice (generated by a Natural-Language-Processing algorithm) affects (dis)honesty, compare it to equivalent human advice, and test whether transparency about advice source matters. We find that dishonesty-promoting advice increases dishonesty, whereas honesty-promoting advice does not increase honesty. This is the case for both AI- and human advice. Algorithmic transparency, a commonly proposed policy to mitigate AI risks, does not affect behaviour. The findings mark the first steps towards managing AI advice responsibly.

U2 - 10.1093/ej/uead056

DO - 10.1093/ej/uead056

M3 - Article

SN - 0013-0133

VL - uead056

JO - Economic Journal

JF - Economic Journal

M1 - uead056

ER -

Corrupted by algorithms? How AI-generated and human-written advice shape (dis)honesty

Abstract

Access to Document

Fingerprint

Cite this